MMP-2 geno-phenotype is prognostic for colorectal cancer survival, whereas MMP-9 is not

The prognostic significance of single-nucleotide polymorphisms (SNPs) and tumour protein levels of MMP-2 and MMP-9 was evaluated in 215 colorectal cancer patients. Single-nucleotide polymorphism MMP-2−1306T and high MMP-2 levels were significantly associated with worse survival. Extreme tumour MMP-9 levels were associated with poor prognosis but SNP MMP-9−1562C>T was not. Tumour MMP levels were not determined by their SNP genotypes.


Patients and study design
Representative, nonnecrotic samples of cancer tissue were collected from 215 patients with colorectal cancer, operated in the Leiden University Medical Centre (December 1983-September 1991. Tissues were snap-frozen and stored at À701C. Clinical data and follow-up were available for at least 10 years. Macroscopic and microscopic parameters were obtained from the pathology reports. The study was performed according to the instructions and guidelines of the LUMC Medical Ethics Committee.

Tissue preparation and protein concentration
Tissues were homogenised in 0.1 M. Tris-HCL (pH 7.5) with 0.1% (v/v) Tween 80 buffer and centrifuged twice all at 41C, as described before (Sier et al, 1991), the protein concentrations were determined (Lowry et al, 1951) and the supernatants were aliquoted and stored at À701C. Storage-induced degradation of MMP-2 and MMP-9 was checked by western blots and gelatin zymography prior to ELISA measurements (Sier et al, 1996).

Determination of MMP-2 and MMP-9 in tissue homogenates
The MMP-2 and MMP-9 levels were determined by previously described ELISAs (Gao et al, 2005). Polyclonal anti-MMP-2 or monoclonal anti-MMP-9 antibodies were used as catching antibody, incubated with appropriately diluted samples (o/n, 41C), and polyclonal anti-MMP-2/biotin-labelled goat anti-rabbit-IgG and biotin-labelled polyclonal anti-MMP-9 antibodies for immunedetection combined with avidin-peroxidase and the 3,3 0 ,5,5 0tetramethyl benzidine/H 2 O 2 substrate solution, and the absorption was read at 450 nm. Sample MMP concentrations were calculated from standard curves and expressed in ng per mg protein.

Single-nucleotide polymorphism analysis
Tissue DNA was isolated using the salting out method (Miller et al, 1988). Single-nucleotide polymorphism (SNP) analysis for MMP-2 À1306C4T and MMP-9 À1562C4T was performed by tetraprimer ARMS PCR, involving four oligonucleotide primers but no restriction enzymes, or RFLP-PCR, as described earlier (Meijer et al, 2006;Kubben et al, 2006a).

Statistical analysis
Statistical analyses were performed using SPSS 12.0 Statistical Package (2004; SPSS Inc., Chicago, IL, USA). Expression differences between groups were calculated using the Mann -Whitney's U-test. Log-rank statistics was used for optimal cutoff point analysis. Hardy -Weinberg's analysis was performed using w 2 or Fisher's exact test to examine differences in the distribution of alleles and genotypes. Correlations between parameters were according to Pearson or Spearman, where appropriate. Overall survival curves were according to Kaplan and Meier. Univariate and multivariate survival analyses were performed using the Cox's proportional hazards method. P-values p0.05 were considered significant.

RESULTS
The genotype distributions of the SNPs for MMP-2 and MMP-9 in 215 colorectal cancer patients are shown in Table 1. The distribution of the polymorphisms in the patients was according to the predicted Hardy -Weinberg's distribution. Both SNPs showed a weak but significant association with TNM stage (Table 1) and MMP-2 À1306C4T also with survival ( Figure 1A). All other clinicopathological parameters did not show an association with either SNP MMP-2 À1306C4T or MMP-9 À1562C4T , and the latter was also not associated with survival ( Figure 1B).
The protein levels of MMP-2 (median 10.6; range: 0.0 -76.6 ng per mg protein) and MMP-9 (median 37.3; range: 0.5 -201.9 ng per mg protein) in the colorectal carcinomas were found not to be related to any of the clinicopathological parameters, although for MMP-2 a stepwise increase with TNM stage was discernable (not shown). For the survival analysis, an optimised cutoff point value was identified for the tumour MMP-2 level (18.5 ng per mg protein, LR 5.07, P ¼ 0.024, Figure 2A). The same approach for MMP-9 resulted in two differently oriented cutoff points; a low MMP-9 value (11.2 ng per mg protein, LR 9.18, P ¼ 0.010) and a high value (125.0 ng per mg protein, LR 5.31, P ¼ 0.021), both associated with poor prognosis ( Figure 2B).
The MMP-2 and MMP-9 protein levels in the colorectal cancer homogenates did not correlate with their respective SNP genotypes. The median values for MMP-2 À1306C4T were 10.9, 9.5 and 11.6 ng mg À1 for genotypes CC, CT and TT, respectively. Even an apparent enhancement of MMP-9 protein associated with the TT genotype of SNP MMP-9 À1562C4T in tumours did not reach significance (CC/CT vs TT, medians 36.5/26.3 vs 44.4 ng mg À1 , Univariate Cox's survival analyses confirmed the association of the MMP-2 SNP and the protein levels of MMP-2 and MMP-9 with survival (Table 2). Multivariate analysis against the prognosisassociated parameters gender, age and TNM classification showed that the MMP-2 SNP was independently associated with survival, whereas the tumour protein levels of MMP-2 just lost and MMP-9 completely lost their significance.

DISCUSSION
MMP-2 and MMP-9 are proteinases implicated in cancer progression. We showed previously that high tumour levels of MMP-2 in gastric carcinomas were consistently associated with a worse survival (Sier et al, 1996;Kubben et al, 2006b). For enhanced MMP-9 levels, the relation with survival was more ambivalent (Kubben et al, 2006b(Kubben et al, , 2007. Also in our present cohort of colorectal carcinoma patients, we found that MMP-2 and MMP-9 levels within the tumours are of significance to survival. The cutoff point analysis showed a broad range of MMP-2 levels with a significant and unidirectional relation with survival outcome: high tumour MMP-2 levels are unfavourable for the patients' prognosis. Our ELISA-derived MMP-2 data correspond very well with a recent immunohistochemical study in a group of 351 colorectal cancer patients, showing that high expression of MMP-2 in malignant epithelium as well as in the surrounding stroma was associated with reduced survival (Hilska et al, 2007). Similar analysis on our tumour MMP-9 data revealed that not only patients with the highest, but also those with the lowest levels had a worse survival  than patients with intermediate levels. Duality of tumour MMP-9 levels with respect to survival has been recognised before in immunohistochemistry-based studies. For example, extensive MMP-9 staining in ovarian cancer cells was associated with a longer survival, as opposed to a shorter survival with a higher stromal expression of MMP-9 (Sillanpaa et al, 2007). Earlier morphometric studies indicated that the degree of MMP-9 expression in tumour-associated lymphocytes, macrophages and neutrophils was inversely associated with invasion and metastasis in colorectal cancer (Takeha et al, 1997). These observations are relevant to our findings because a low level of MMP-9 in the tumour homogenates might indicate a lack of infiltration of the tumours with MMP-9 containing leukocytes, known to exert anticancer effects (Nielsen et al, 1996;Owen et al, 2004), leading to an adverse prognosis. In addition, MMP-9-mediated cleavage of extracellular matrix components is also known to generate antiangiogenesis inhibitors like angiostatin (Patterson and Sang, 1997). Low MMP-9 expression in the tumour, leading to insufficient production of antiangiogenic factors, could also contribute to the worse prognosis of these patients (Pozzi et al, 2002). However, MMP-9 has a wide range of substrates, including various growth factors and several types of collagen,  which after cleavage contribute to the process of invasion, angiogenesis and metastasis of tumours, explaining why high tumour levels of MMP-9, in general, are associated with a poor prognosis. Differences in expression of metalloproteinases might, in part, be explained by genotypic variation. Because the investigated SNPs of MMP-2 and MMP-9 are located in the promoter region of the gene, a correlation between polymorphism and protein expression might be expected. The C-T transition at the À1306 position of the MMP-2 gene promoter prevents binding of the stimulating Sp1 transcription factor, whereas a change at the À1562 position of the MMP-9 gene decreases binding of a repression factor (Zhang et al, 1999;Price et al, 2001). In vitro MMP-2 expression levels by colon cancer cell lines containing the CC genotype were indeed higher compared with cells with the CT genotype (Xu et al, 2004). In our cohort of patients, however, we found no relationship between MMP-2 and MMP-9 polymorphisms and the tumour protein levels, likely caused by the complicated regulatory posttranslational mechanisms for proteinases in multicellular tumour tissues. Nevertheless, we did find significant associations between MMP-2 À1306C4T and TNM stage and survival, that is, worse prognosis in patients with the TT genotype. This association with the outcome of the patients support the recent confirmation of MMP-2 as one of the candidate cancer genes (CAN genes) by the number and nature of mutations and pathways in colorectal cancer (Sjöblom et al, 2006). Our results for SNP MMP-9 À1562C4T correspond well with data from Asiatic patients (Xu et al, 2007), indicating that this polymorphism is not directly involved in the process of colorectal carcinogenesis.
Summarising, we showed that MMP-9, mechanistically probably the most interesting of the gelatinases, is not a likely candidate as a simple prognostic indicator. Despite the absence of a correlation between promoter-located SNP À1306C4T in the MMP-2 gene with tumour MMP-2 levels, both parameters were significantly associated with survival, indicating MMP-2 as a consistent independent prognostic factor in colorectal cancer.